mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-11-14 00:26:28 +01:00
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - The chat-feature evals measure whether agents complete real chat requests through the product > - The chat catalog eval harness and its repairs live on the `chat-eval/master-candidate` branch, not on master > - Today's work made 139 of 242 chat evals pass in verified single-case live runs. Each product fix landed on master as its own pull request > - The eval-side changes (setups, graders, fault seams, evidence capture) still exist only on the candidate > - This pull request lands the harness on master. Master stays the source of truth for product behavior > - The benefit is one tree: evals run against master commits, and product fixes reach the evals as soon as they merge ## Linked Issues or Issue Description Refs #19441, Refs #19599, Refs #20270, Refs #20334, Refs #20342, Refs #20344, Refs #20354, Refs #20360, Refs #20373, Refs #20382, Refs #20409 **What existing behavior does this improve?** The chat catalog eval harness and the product it evaluates live on a long-running branch outside master. Eval fixes do not reach master, so master-based eval runs use old setups and graders. **Subsystem affected** `tests/runner-e2e` chat catalog harness, plus a small set of product changes that exist only on the candidate. **Proposed change** Land the harness on master. Every product fix from this eval work already merged to master as its own pull request. Conflicts keep master's product behavior. ## What Changed - Most of the diff (808 of 851 files) is under `tests/runner-e2e/`: chat-catalog domains, scenarios, descriptors, the TestOps manifest, private evidence tooling, and their tests. - Graders now check saved state, setup literals and rendered or clicked links. Many keyword and regex checks on free-form replies are removed. Each changed grader has a bumped `oracleVersion` and calibration tests (a valid reply, a reworded valid reply, and broken variants that must fail). - The shared link helper `hasLink` no longer treats a list link (for example `/COM/issues?view=mine`) as a link to every task without a URL key. - Setup fixes build real state through supported product routes: routine history, agent grants through the board grant route, budget holds with a real cap, and a storage-layer archive fault for `inbox-search.archive-partial-failure`. - Browser boundary journeys (sign-in, chat-disabled, onboarding) now write the standard pass evidence files. - Retargeted cases, approved by the maintainer: `agent-chat.chat-disabled` (Conference Room gate), `agents.restore-instructions` (no saved instruction history), `access.manage-membership` (member permission edits stay board-only). - The private evidence trusted files (`manual-private-handoff.ts`, `manual-private-retention.ts`, `manual-private-report-assets.ts`) are byte-identical to master. - Product files outside `tests/` (43 files, +1,328 −49) carry candidate-only changes that master does not yet have. Most are tests. The main code changes are in `request-human-input.ts`, `heartbeat-run-summary.ts`, `native-completion-feedback.ts`, `native-run-finalizer.ts`, `openapi.ts`, `secrets.ts` and `IssueChatLongThreadPerf.tsx`. ## Verification - `pnpm vitest run chat-catalog-domains chat-catalog-contract.test.ts chat-catalog-consistency.test.ts unit-launcher.test.mjs --testTimeout=180000` in `tests/runner-e2e`: 269 files pass, 3,550 tests pass. - `node --test chat-full-catalog-testops-manifest.test.mjs`: 10 pass. - `node tests/runner-e2e/source-sdk-public-inputs.mjs <HEAD> --check`: no change. - `tsc` for `server` and `tests/runner-e2e`: clean. - Live single-case verification: 139 of 242 cases passed on paid runs, and a person read the saved evidence for each. These are single-case runs, not one full-cohort run. ## Risks - Large diff. Most of it is test-only, but the 43 non-test files change product code paths. Review those files first. - Some graders are now less strict on reply wording. A person must read replies for meaning on those cases. - 42 cases need disposable outside accounts (Notion, Slack, GitHub, Daytona and others) and stay blocked. `access.invite-teammate` stays failing until local agent isolation lands, because the runner's durable state keeps a live invite token. - The paid workflow compares the trusted private-evidence files with master. They are unchanged, so this risk is low. ## Model Used Anthropic Claude Opus 5.5 (`claude-opus-5-5`), with tool use and code execution, in Claude Code. Subagents used the same model. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: Paperclip <noreply@paperclip.ing>
109 lines
9.7 KiB
TypeScript
109 lines
9.7 KiB
TypeScript
import { expect } from "@playwright/test";
|
|
import { createHash } from "node:crypto";
|
|
import { pollUntil } from "../api.js";
|
|
import { dropChatSendAcknowledgement } from "../lost-send.js";
|
|
import type { ChatQueryAttachmentReceipt } from "../chat-query-attachments.js";
|
|
import { checks, sameTask, type Entity } from "./internal-support.js";
|
|
import { openStandingChat } from "./identity-support.js";
|
|
import { runtimeDefinition, runtimeWorker, runtimeObservation, observeRuntime } from "./runtime-support.js";
|
|
import type { ChatCatalogContext } from "../chat-catalog-contract.js";
|
|
type Interception = Awaited<ReturnType<typeof dropChatSendAcknowledgement>>;
|
|
const interceptions = new WeakMap<ChatCatalogContext, Interception>();
|
|
interface DraftComment extends Entity { clientRequestId?: string }
|
|
const editor = (ctx: ChatCatalogContext) => ctx.input.page.getByTestId("task-chat-composer-input").last().locator('[contenteditable="true"], textarea').first();
|
|
async function draftText(ctx: ChatCatalogContext) {
|
|
return editor(ctx).evaluate(node => "value" in node ? String(node.value) : node.textContent ?? "");
|
|
}
|
|
export function gradePreserveDraft(value: unknown) {
|
|
const c = checks(); const o = runtimeObservation(value);
|
|
c.add("real-draft-and-upload-evidence", !!o, "The unsent editor text, completed upload, source comment and actual transport receipt are independently required."); if (!o) return c.result();
|
|
const attachment = o.state.briefAttachment as ChatQueryAttachmentReceipt;
|
|
const sourceComments = o.state.sourceComments as DraftComment[] | undefined;
|
|
const transport = o.state.transport as { commentId?: string; replayedCommentId?: string; clientRequestId?: string } | undefined;
|
|
const human = sourceComments?.filter(row => !row.authorAgentId);
|
|
c.add("draft-survived-switch-and-reload", o.state.restoredDraft === o.state.originalQuery && o.state.restoredAttachmentVisible === true
|
|
&& o.state.beforeSendHumanCount === 0 && o.state.otherTaskDraftIsolated === true && o.state.otherTaskAttachmentsEmpty === true,
|
|
"The correct source retains the exact unsent text and completed upload while the other task receives neither.");
|
|
const canonical = human?.find(row => row.id === transport?.commentId);
|
|
const savedAttachment = o.state.savedAttachment as ChatQueryAttachmentReceipt | undefined;
|
|
c.add("one-message-with-original-upload", human?.length === 1 && !!canonical && !!attachment && savedAttachment?.id === attachment.id
|
|
&& savedAttachment.issueCommentId === canonical.id && savedAttachment.issueId === o.state.sourceIssueId
|
|
&& savedAttachment.companyId === o.companyId && savedAttachment.createdByUserId === canonical.authorUserId
|
|
&& savedAttachment.createdByAgentId == null && savedAttachment.sha256 === o.state.briefHash,
|
|
"One actual human send carries the exact human-owned upload receipt for this source and company.");
|
|
c.add("actual-file-bytes-retained", o.state.downloadedBrief === o.state.briefBody && typeof o.state.briefBody === "string"
|
|
&& createHash("sha256").update(o.state.briefBody).digest("hex") === savedAttachment?.sha256,
|
|
"Independent production download bytes match the original brief and receipt hash.");
|
|
const requestRuns = transport ? o.runs.filter(run => run.contextSnapshot?.wakeCommentId === transport.commentId
|
|
|| Array.isArray(run.contextSnapshot?.wakeCommentIds) && run.contextSnapshot.wakeCommentIds.includes(transport.commentId)) : [];
|
|
c.add("uncertain-send-reconciled-without-duplication", !!transport?.commentId && transport.commentId === transport.replayedCommentId
|
|
&& !!transport.clientRequestId && sourceComments?.filter(row => row.clientRequestId === transport.clientRequestId).map(row => row.id).join() === transport.commentId
|
|
&& requestRuns.length === 1 && o.state.persistedCommentReviewedBeforeReplay === true,
|
|
"A real lost acknowledgement is reconciled against the persisted comment before an exact client-request replay; one request-bound run remains.");
|
|
c.add("other-task-unchanged", !!o.state.otherTask && sameTask(o.state.otherTask as Entity, o.tasks.find(row => row.id === (o.state.otherTask as Entity).id)),
|
|
"Draft navigation and the later send do not mutate the unrelated task.");
|
|
return c.result();
|
|
}
|
|
export const runtimeSteeringCases = [runtimeDefinition("steering.preserve-draft", async ctx => {
|
|
const worker = await runtimeWorker(ctx); const source = await openStandingChat(ctx);
|
|
await ctx.input.page.reload({ waitUntil: "domcontentloaded" }); await expect(editor(ctx)).toBeVisible();
|
|
const otherTask = await ctx.input.api.post<Entity>(`/api/companies/${ctx.input.fixtures.company.id}/issues`, { title: "Unrelated draft-isolation task", status: "backlog" });
|
|
const sourceUrl = ctx.input.page.url(); const filename = "website-brief.md";
|
|
const briefBody = `# Website brief\nUse short copy, an accessible single-page layout, and a contact section. Product reference: ${ctx.input.nonce}.\n`;
|
|
await editor(ctx).fill(ctx.scenario.query);
|
|
await ctx.input.page.locator('input[type="file"]').first().setInputFiles({ name: filename, mimeType: "text/markdown", buffer: Buffer.from(briefBody) });
|
|
const briefAttachment = await pollUntil({ label: "Completed public brief upload before unsent navigation", deadlineAt: Date.now() + 30_000,
|
|
load: () => ctx.input.api.get<ChatQueryAttachmentReceipt[]>(`/api/issues/${source.id}/attachments`),
|
|
accept: rows => rows.filter(row => row.originalFilename === filename).length === 1 });
|
|
const receipt = briefAttachment.find(row => row.originalFilename === filename)!;
|
|
await expect(ctx.input.page.getByTestId("task-chat-composer-attachments")).toContainText(filename);
|
|
await expect(ctx.input.page.getByTestId("task-chat-composer-attachments")).not.toContainText("Uploading");
|
|
await ctx.input.page.goto(`/${ctx.input.fixtures.company.issuePrefix}/issues/${otherTask.identifier ?? otherTask.id}`, { waitUntil: "domcontentloaded" });
|
|
await expect(editor(ctx)).toBeVisible(); await ctx.input.page.reload({ waitUntil: "domcontentloaded" });
|
|
await expect(editor(ctx)).toBeVisible();
|
|
const otherTaskDraftIsolated = (await draftText(ctx)) !== ctx.scenario.query;
|
|
const otherTaskAttachmentsEmpty = (await ctx.input.api.get<Entity[]>(`/api/issues/${otherTask.id}/attachments`)).length === 0
|
|
&& !(await ctx.input.page.getByTestId("task-chat-composer-attachments").isVisible());
|
|
await ctx.input.page.goto(sourceUrl, { waitUntil: "domcontentloaded" });
|
|
await ctx.input.page.reload({ waitUntil: "domcontentloaded" }); await expect(editor(ctx)).toBeVisible();
|
|
await expect.poll(() => draftText(ctx)).toBe(ctx.scenario.query);
|
|
const restoredDraft = await draftText(ctx);
|
|
await expect(ctx.input.page.getByTestId("task-chat-composer-attachments")).toContainText(filename);
|
|
const before = await ctx.input.api.get<Entity[]>(`/api/issues/${source.id}/comments?order=asc`);
|
|
interceptions.set(ctx, await dropChatSendAcknowledgement(ctx.input.page, ctx.scenario.query));
|
|
return { worker, otherTask, sourceIssueId: source.id, originalQuery: ctx.scenario.query, restoredDraft,
|
|
restoredAttachmentVisible: true, otherTaskDraftIsolated, otherTaskAttachmentsEmpty, beforeSendHumanCount: before.filter(row => !row.authorAgentId).length,
|
|
briefAttachment: receipt, briefBody, briefHash: createHash("sha256").update(briefBody).digest("hex") };
|
|
}, gradePreserveDraft, {
|
|
authentication: "human-session", queryAttachments: [{ stateKey: "briefAttachment", filename: "website-brief.md" }],
|
|
sourceFiles: ["chat-catalog-domains/runtime-steering.ts", "chat-catalog-domains/runtime-support.ts", "lost-send.ts", "chat-scenarios/steering.yaml"],
|
|
expectedRunCount: 3, minimumRunCount: 1, attemptTimeoutMs: 540_000,
|
|
runPolicy: { kind: "source-settled-with-active-work", maximumActiveWorkerRuns: 1 },
|
|
async act(ctx, state) {
|
|
const interception = interceptions.get(ctx); if (!interception) throw new Error("Lost-ack draft fixture is not armed");
|
|
const sent = await interception.committed;
|
|
const comment = await ctx.input.api.get<DraftComment>(`/api/issues/${state.sourceIssueId}/comments/${sent.commentId}`);
|
|
if (comment.id !== sent.commentId || comment.clientRequestId !== sent.data.clientRequestId) throw new Error("Uncertain send was not reconciled with its actual persisted comment");
|
|
state.persistedCommentReviewedBeforeReplay = true;
|
|
await interception.dispose(); interceptions.delete(ctx);
|
|
const replayed = await ctx.input.api.post<{ id: string }>(sent.path, sent.data);
|
|
state.transport = { commentId: sent.commentId, replayedCommentId: replayed.id, clientRequestId: String(sent.data.clientRequestId) };
|
|
await ctx.input.page.reload({ waitUntil: "domcontentloaded" });
|
|
},
|
|
async observe(ctx, state) {
|
|
const source = String(state.sourceIssueId);
|
|
state.sourceComments = await ctx.input.api.get<DraftComment[]>(`/api/issues/${source}/comments?order=asc`);
|
|
const attachment = (await ctx.input.api.get<ChatQueryAttachmentReceipt[]>(`/api/issues/${source}/attachments`)).find(row => row.id === (state.briefAttachment as ChatQueryAttachmentReceipt).id);
|
|
state.savedAttachment = attachment;
|
|
if (attachment) state.downloadedBrief = Buffer.from(await ctx.input.api.getAttachmentContent(attachment.id)).toString("utf8");
|
|
return observeRuntime(ctx, state);
|
|
},
|
|
async verifyResult(ctx, state) {
|
|
const transport = state.transport as { commentId: string }; const receipt = state.savedAttachment as ChatQueryAttachmentReceipt;
|
|
const rendered = ctx.input.page.locator(`[id=${JSON.stringify(`comment-${transport.commentId}`)}]`);
|
|
await expect(rendered).toContainText(ctx.scenario.query);
|
|
await expect(rendered.locator(`a[href=${JSON.stringify(receipt.contentPath)}]`).first()).toBeVisible();
|
|
},
|
|
async cleanup(ctx) { const interception = interceptions.get(ctx); interceptions.delete(ctx); await interception?.dispose(); },
|
|
})];
|